Autophagy is a conserved pathway for cell survival during starvation and for clearance of damaged, toxic, or excess organelles and inclusions from the cell. Bulk cytosol, organelles, and other autophagic cargo are taken up within a double membrane vesicle known as the autophagosome. Autophagy is central to cell function and human health, yet the physical basis of autophagosome biogenesis is almost completely unknown. The proposed research will yield, in near-atomistic detail, the mechanism for the earliest stage in autophagosome biogenesis. We hypothesize that the key event in the initiation of the double membrane of the autophagosome is the clustering of high curvature vesicles containing the membrane protein Atg9 at the preautophagosomal structure (PAS). This event is orchestrated by the Atg1 complex, which in yeast consists of Atg1, Atg13, Atg17, Atg29, and Atg31. The latter three comprise the Atg17-Atg31-Atg29 subcomplex, which is the first to arrive at the PAS. We determined the structure of the Atg17-Atg31-Atg29 complex, revealing a remarkable S-shaped double crescent and suggesting a model for vesicle scaffolding. We also discovered that the C-terminal early autophagy targeting/tethering (EAT) domain of Atg1 is a potent sensor for high membrane curvature and tethers high curvature vesicles. These insights led us to a detailed hypothesis for the scaffolding and tethering of the vesicles that initiate autophagosome biogenesis. The specific aims of this project are as follows: 1. we will understand how the Atg1 EAT domain tethers highly curved vesicles. Tethering will be assayed in vitro and correlated with function in yeast cells. The structural basis for tethering will be determined using scanning mutagenesis and hydrogen-deuterium exchange coupled to mass spectrometry. 2. The three dimensional structure of the Atg1 complex will be determined by SAXS and EM, and structural interfaces mapped by scanning mutagenesis to develop a pseudo-atomic model for the complex. 3. In a stringent test of the central hypothesis, an Atg9 lipopeptide based model for precursor vesicles will be used to reconstitute the early PAS in vitro. The model system will be used to probe whether Atg13 dephosphorylation during starvation triggers the assembly of a tethering-competent form of the Atg1 complex. Together, these aims will flesh out the main early events of autophagy initiation in mechanistic and structural detail.